Matlab 去趋势处理

1.介绍

去趋势处理可以去除传感器获取数据时产生的偏移，可以将数据分析集中在波动上

2.原理

对数据减去一条最优（拟合）的直线，平面或者曲面，使得处理后的数据均值为零

3.函数

3.1连续线性趋势

 y  = detrend(x)     # 从x的数据中去除最优的直线拟合

 y = detrend(x, n)  # 当n = 0 , 去除均值；n = 1, x去除线性趋势；n = 2 ，去除二次线性趋势

clc;
clear all;
close all;

t = 0:20
x = 3*sin(t) + t
y = detrend(x)
y1 = detrend(x,2)

figure(1)
plot(t, x, t, y, t, x-y, "k:")
legend("input data", "Detrend data", "trend","location","northwest")

3.2 最小二乘法拟合消除

[y,xtrend]=polydetrend(x,fs,m)  # fs:采样频率； m: 多项式阶数； y:去趋势的结果； xtrend: 拟合的曲线  


clc;
clear all;
close all;


t = 0:20
x = 3*sin(t) + t
y = detrend(x)
[y1,xtrend] = polydetrend(x,100,6);


figure(1)
plot(t, x, t, y, t, x-y, "k:")
legend("input data", "Detrend data", "trend","location","northwest")

figure(2)
plot(t, x, t, y1, t, xtrend, "k:")
legend("input data", "Detrend data", "trend","location","northwest")



function [y,xtrend] = polydetrend(x,fs,m)
    x = x(:);                % 把输入信号x转换为列数据
    N = length(x);           % 获取x的长度N
    t = (0:N-1)'/fs;         % 按x的长度和采样频率设置时间序列
    a = polyfit(t,x,m);      % 用最小二乘法拟合语音信号x的多项式系数a
    xtrend = polyval(a,t);   % 用系数a和时间序列t构成趋势项
    y = x-xtrend;            % 从信号x中清楚趋势项 
end